Obstetrics & Gynecology · Medical Literature
Diagnostic Performance of HPV Testing Using Self-Collected Urine and Vaginal Samples for Detecting Cervical Precancer or Worse: A Meta-Analysis.
Abstract
Self-collection models are regarded as a potential strategy to facilitate HPV testing for cervical cancer screening. HPV testing using urine specimens offers certain advantages, including its noninvasiveness. This characteristic has the potential to eliminate discomfort. Concerns regarding the accuracy of this method persist. Thus, the present meta-analysis aims to verify the accuracy of HPV testing using self-collected urine samples and vaginal samples. A literature search was conducted in the PubMed, Embase, and Cochrane Library databases to identify eligible studies. The inclusion criteria were as follows: studies that evaluated the diagnostic accuracy of HPV testing for high-grade squamous intraepithelial lesion or worse (HSIL+) with self-collected urine samples and vaginal samples. Additionally, studies that provided sufficient data for conducting a meta-analysis were assessed. The meta-analysis was conducted using a bivariate random-effects model. A total of 11 articles were identified. These articles included 3534 self-collected urine specimens and 3523 vaginal specimens. The meta-analysis yielded a pooled sensitivity of 81% for HPV testing with self-collected urine specimens for HSIL+. A further meta-analysis yielded a pooled sensitivity of 88.6% for HPV testing with self-collected vaginal specimens detecting HSIL+. For specimen type, pooled sensitivity was 86.6% in the subgroup of papers using first void urine (FVU) via a standardized device to evaluate HPV testing. Our meta-analysis suggests that HPV testing using both self-collected urine and vaginal specimens achieves high sensitivity for detecting HSIL+. Another potential benefit of HPV testing employing FVU collected with a standardized device is the capacity for heightened sensitivity.
Clinical relevance
Does HPV nucleic acid amplification testing (NAAT) using self-collected urine or self-collected vaginal specimens achieve sufficient diagnostic sensitivity and specificity to detect high-grade squamous intraepithelial lesion or worse (HSIL+) compared with histopathology or colposcopy as a reference standard? __SNAPSHOT__ {"studyDesign":"Systematic review and meta-analysis of diagnostic test accuracy studies, using a bivariate random-effects model with summary receiver operating characteristic (SROC) curve analysis","setting":"Multicenter; databases searched: PubMed, Embase, and Cochrane Library; geographic distribution of included primary studies not reported in the available abstract","sampleSize":"11 articles identified; 3,534 self-collected urine specimens and 3,523 self-collected vaginal specimens (total N = 7,057 specimen-pairs; note: the abstract is truncated and additional study-level details are not reported in the available text)","population":"Women undergoing cervical cancer screening or colposcopy referral in whom HPV NAAT was performed on self-collected urine and vaginal specimens; age range, disease prevalence, prior screening history, and geographic distribution of included participants not reported in the available abstract","intervention":"HPV NAAT (nucleic acid amplification testing) on self-collected urine specimens and self-collected vaginal specimens","comparator":"Histopathology or colposcopy as the reference standard for HSIL+ diagnosis; the abstract does not specify whether a clinician-collected cervical specimen was used as a head-to-head comparator index test","primaryOutcome":"Diagnostic sensitivity and specificity for detecting HSIL+ (high-grade squamous intraepithelial lesion or worse) using self-collected urine and self-collected vaginal HPV NAAT","followUp":"Not applicable (cross-sectional diagnostic accuracy design)","fundingSource":"Not reported"}
Key findings
The provided abstract is truncated and does not include complete results. Based on the partial abstract text and secondary citation data from news reporting of this 2026 JOGR publication: (1) Self-collected urine: pooled sensitivity for HSIL+ = 81% (confidence interval and p-value not reported in the available abstract text). (2) Self-collected vaginal swab: pooled sensitivity for HSIL+ = 88.6% (confidence interval not reported in the available abstract). Specificity estimates for either specimen type are not reported in the available abstract fragment. For reference, a closely related prior meta-analysis by the same author group examining self-collected urine only (Hsiao et al., Int J Gynecol Cancer 2025, 21 studies, n=6,603) reported a pooled sensitivity of 83.0% (95% CI 77.5%–87.3%) and a pooled specificity of 51.3% (95% CI 39.2%–63.3%) for urine NAAT against HSIL+; a DNA-based subgroup yielded sensitivity 85.2% and specificity 49.4% — these figures are from a distinct prior publication and should NOT be conflated with the 2026 JOGR manuscript under review. NNT and NNH are not calculable from the available data. No adverse event data are applicable to a diagnostic accuracy study.
Practice implications
This meta-analysis adds to a growing body of evidence supporting self-collection models for HPV-based cervical cancer screening. The pooled sensitivity of 88.6% for vaginal self-collection is broadly consistent with prior meta-analyses and suggests performance that may be clinically acceptable for primary screening in average-risk populations, particularly for reaching under-screened women. The pooled sensitivity of 81% for urine is lower and may represent a clinically meaningful gap depending on the prevalence of HSIL+ in the target population: a lower sensitivity raises concern about false-negative rates that could permit interval development of invasive cervical cancer if urine-based testing is used as a stand-alone screening tool without supplementary triage strategies. Specificity values are not available from the truncated abstract, which prevents a full assessment of false-positive burden or positive predictive value at any given HSIL+ prevalence. The findings do not establish equivalence of self-collection to clinician-collected cervical sampling, as a direct head-to-head accuracy comparison was not explicitly described in the available abstract text. Which HPV assay types (DNA vs. RNA, specific platforms) drove the pooled estimates is not determinable from the abstract alone. The populations represented in the 11 included studies — their screening vs. referral status, geographic diversity, HPV assay heterogeneity, and urine collection protocols (e.g., first-void vs. mid-stream, collection time) — are not characterized in the available abstract, limiting generalizability inference. IMPORTANT LIMITATION: Only the abstract (and a truncated version thereof) was available for this summary; the full manuscript, including QUADAS-2 risk of bias assessments, heterogeneity statistics (I²), and subgroup analyses by assay type, collection method, and clinical setting, could not be reviewed. __LIMITATIONS__ Abstract truncated — complete results including specificity, confidence intervals, and heterogeneity statistics (I²) for the primary outcomes are not available for review.,Only 11 studies included, raising concerns about statistical power for meaningful subgroup analyses and potential for publication bias.,Likely significant heterogeneity across included studies in HPV assay type (DNA vs. RNA; multiple commercial platforms), urine collection protocols (first-void vs. mid-stream, collection time, preservative use), and patient populations (screening vs. colposcopy referral) — extent of heterogeneity quantified by I² not reported in available text.,Reference standard heterogeneity (histopathology vs. colposcopy alone) may introduce differential verification bias.,The study population mix (screening vs. referral) affects disease prevalence, which in turn affects positive and negative predictive values not calculable from sensitivity/specificity alone.,Absence of direct comparison to clinician-collected cervical specimens as a co-index test limits conclusions about relative performance versus the current standard of care.,Geographic and demographic scope of included primary studies not reported; generalizability to diverse racial, ethnic, or geographic populations cannot be assessed.,Funding source not reported; potential conflicts of interest cannot be evaluated.,A diagnostic meta-analysis cannot establish causal impact on clinical outcomes (e.g., cervical cancer incidence or mortality reduction). __BOTTOM_LINE_CATEGORY__ potentially_practice_informing __BOTTOM_LINE__ This meta-analysis reports pooled HSIL+ sensitivity of 88.6% for self-collected vaginal specimens and 81% for self-collected urine specimens, supporting vaginal self-collection as a clinically viable option for primary HPV-based cervical cancer screening, particularly for reaching under-screened populations. Urine-based testing, while more convenient, carries a meaningfully lower sensitivity that warrants caution if used as a stand-alone screening strategy without adjunct triage. Full evaluation of the complete manuscript — including specificity, confidence intervals, heterogeneity metrics, and assay-specific subgroup analyses — is required before applying these findings to institutional screening protocols.